X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=lexer.c;h=20993600129097ca06d55f4c1c57611c591ddec8;hb=9de06381987e3b71a9efea26cb9b596fd433191c;hp=cbdb78387af45d2af0484450ec5490a08d3738de;hpb=2a121a9dd08c4edc58cd501838376af6f7d14877;p=cparser diff --git a/lexer.c b/lexer.c index cbdb783..2099360 100644 --- a/lexer.c +++ b/lexer.c @@ -6,15 +6,33 @@ #include "adt/error.h" #include "adt/strset.h" #include "adt/util.h" +#include "type_t.h" #include #include #include +#include #include //#define DEBUG_CHARS #define MAX_PUTBACK 3 +#ifdef _WIN32 +/* No strtold on windows and no replacement yet */ +#define strtold(s, e) strtod(s, e) +#endif + +#define HAS_SIGNED_CHAR +//#define HAS_UNSIGNED_CHAR + +#if defined HAS_SIGNED_CHAR +typedef signed char char_type; +#elif defined HAS_UNSIGNED_CHAR +typedef unsigned char char_type; +#else +# error signedness of char not determined +#endif + static int c; token_t lexer_token; symbol_t *symbol_L; @@ -24,6 +42,15 @@ static const char *bufend; static const char *bufpos; static strset_t stringset; +static type_t *type_int = NULL; +static type_t *type_uint = NULL; +static type_t *type_ulong = NULL; +static type_t *type_longlong = NULL; +static type_t *type_ulonglong = NULL; +static type_t *type_float = NULL; +static type_t *type_double = NULL; +static type_t *type_longdouble = NULL; + static void error_prefix_at(const char *input_name, unsigned linenr) { fprintf(stderr, "%s:%u: Error: ", input_name, linenr); @@ -60,10 +87,10 @@ static inline void next_real_char(void) static inline void put_back(int pc) { assert(bufpos >= buf); - assert(bufpos < buf+MAX_PUTBACK || *bufpos == pc); + //assert(bufpos < buf+MAX_PUTBACK || *bufpos == pc); char *p = buf + (bufpos - buf); - *p = pc; + *p = (char) pc; /* going backwards in the buffer is legal as long as it's not more often * than MAX_PUTBACK */ @@ -110,7 +137,6 @@ static inline void next_char(void) { next_real_char(); -#if 0 /* filter trigraphs */ if(UNLIKELY(c == '\\')) { maybe_concat_lines(); @@ -145,13 +171,9 @@ static inline void next_char(void) break; } -end_of_next_char: -#endif - (void) maybe_concat_lines; +end_of_next_char:; #ifdef DEBUG_CHARS printf("nchar '%c'\n", c); -#else - ; #endif } @@ -227,14 +249,14 @@ static void parse_symbol(void) symbol_t *symbol; char *string; - obstack_1grow(&symbol_obstack, c); + obstack_1grow(&symbol_obstack, (char) c); next_char(); while(1) { switch(c) { DIGITS SYMBOL_CHARS - obstack_1grow(&symbol_obstack, c); + obstack_1grow(&symbol_obstack, (char) c); next_char(); break; @@ -259,14 +281,18 @@ end_symbol: static void parse_integer_suffix(void) { - if(c == 'U' || c == 'U') { - /* TODO do something with the suffixes... */ + if(c == 'U' || c == 'u') { next_char(); if(c == 'L' || c == 'l') { next_char(); if(c == 'L' || c == 'l') { next_char(); + lexer_token.datatype = type_ulonglong; + } else { + lexer_token.datatype = type_ulong; } + } else { + lexer_token.datatype = type_uint; } } else if(c == 'l' || c == 'L') { next_char(); @@ -274,10 +300,18 @@ static void parse_integer_suffix(void) next_char(); if(c == 'u' || c == 'U') { next_char(); + lexer_token.datatype = type_ulonglong; + } else { + lexer_token.datatype = type_longlong; } } else if(c == 'u' || c == 'U') { next_char(); + lexer_token.datatype = type_ulong; + } else { + lexer_token.datatype = type_int; } + } else { + lexer_token.datatype = type_int; } } @@ -287,13 +321,102 @@ static void parse_floating_suffix(void) /* TODO: do something usefull with the suffixes... */ case 'f': case 'F': + next_char(); + lexer_token.datatype = type_float; + break; case 'l': case 'L': next_char(); + lexer_token.datatype = type_longdouble; + break; + default: + lexer_token.datatype = type_double; + break; + } +} + +/** + * A replacement for strtoull. Only those parts needed for + * our parser are implemented. + */ +static unsigned long long parse_int_string(const char *s, const char **endptr, int base) { + unsigned long long v = 0; + + switch (base) { + case 16: + for (;; ++s) { + /* check for overrun */ + if (v >= 0x1000000000000000ULL) + break; + switch (tolower(*s)) { + case '0': v <<= 4; break; + case '1': v <<= 4; v |= 0x1; break; + case '2': v <<= 4; v |= 0x2; break; + case '3': v <<= 4; v |= 0x3; break; + case '4': v <<= 4; v |= 0x4; break; + case '5': v <<= 4; v |= 0x5; break; + case '6': v <<= 4; v |= 0x6; break; + case '7': v <<= 4; v |= 0x7; break; + case '8': v <<= 4; v |= 0x8; break; + case '9': v <<= 4; v |= 0x9; break; + case 'a': v <<= 4; v |= 0xa; break; + case 'b': v <<= 4; v |= 0xb; break; + case 'c': v <<= 4; v |= 0xc; break; + case 'd': v <<= 4; v |= 0xd; break; + case 'e': v <<= 4; v |= 0xe; break; + case 'f': v <<= 4; v |= 0xf; break; + default: + goto end; + } + } + break; + case 8: + for (;; ++s) { + /* check for overrun */ + if (v >= 0x2000000000000000ULL) + break; + switch (tolower(*s)) { + case '0': v <<= 3; break; + case '1': v <<= 3; v |= 1; break; + case '2': v <<= 3; v |= 2; break; + case '3': v <<= 3; v |= 3; break; + case '4': v <<= 3; v |= 4; break; + case '5': v <<= 3; v |= 5; break; + case '6': v <<= 3; v |= 6; break; + case '7': v <<= 3; v |= 7; break; + default: + goto end; + } + } + break; + case 10: + for (;; ++s) { + /* check for overrun */ + if (v > 0x1999999999999999ULL) + break; + switch (tolower(*s)) { + case '0': v *= 10; break; + case '1': v *= 10; v += 1; break; + case '2': v *= 10; v += 2; break; + case '3': v *= 10; v += 3; break; + case '4': v *= 10; v += 4; break; + case '5': v *= 10; v += 5; break; + case '6': v *= 10; v += 6; break; + case '7': v *= 10; v += 7; break; + case '8': v *= 10; v += 8; break; + case '9': v *= 10; v += 9; break; + default: + goto end; + } + } break; default: + assert(0); break; } +end: + *endptr = s; + return v; } static void parse_number_hex(void) @@ -301,135 +424,117 @@ static void parse_number_hex(void) assert(c == 'x' || c == 'X'); next_char(); - if (!isdigit(c) && - !('A' <= c && c <= 'F') && - !('a' <= c && c <= 'f')) { - parse_error("premature end of hex number literal"); - lexer_token.type = T_ERROR; - return; - } - - int value = 0; - while(1) { - if (isdigit(c)) { - value = 16 * value + c - '0'; - } else if ('A' <= c && c <= 'F') { - value = 16 * value + c - 'A' + 10; - } else if ('a' <= c && c <= 'f') { - value = 16 * value + c - 'a' + 10; - } else { - parse_integer_suffix(); - - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = value; - return; - } + while(isxdigit(c)) { + obstack_1grow(&symbol_obstack, (char) c); next_char(); } + obstack_1grow(&symbol_obstack, '\0'); + char *string = obstack_finish(&symbol_obstack); if(c == '.' || c == 'p' || c == 'P') { next_char(); panic("Hex floating point numbers not implemented yet"); } -} - -static void parse_number_oct(void) -{ - int value = 0; - while(c >= '0' && c <= '7') { - value = 8 * value + c - '0'; - next_char(); - } - if (c == '8' || c == '9') { - parse_error("invalid octal number"); + if(*string == '\0') { + parse_error("invalid hex number"); lexer_token.type = T_ERROR; - return; } + const char *endptr; lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = value; + lexer_token.v.intvalue = parse_int_string(string, &endptr, 16); + if(*endptr != '\0') { + parse_error("hex number literal too long"); + } + obstack_free(&symbol_obstack, string); parse_integer_suffix(); } -static void parse_floatingpoint_exponent(long double value) +static inline bool is_octal_digit(int chr) { - unsigned int expo = 0; - long double factor = 10.; - - if(c == '-') { - next_char(); - factor = 0.1; - } else if(c == '+') { - next_char(); - } - - while(c >= '0' && c <= '9') { - expo = 10 * expo + (c - '0'); - next_char(); - } - - while(1) { - if(expo & 1) - value *= factor; - expo >>= 1; - if(expo == 0) - break; - factor *= factor; - } - - lexer_token.type = T_FLOATINGPOINT; - lexer_token.v.floatvalue = value; - - parse_floating_suffix(); + return '0' <= chr && chr <= '7'; } -static void parse_floatingpoint_fract(int integer_part) +static void parse_number_oct(void) { - long double value = integer_part; - long double factor = 1.; - - while(c >= '0' && c <= '9') { - factor *= 0.1; - value += (c - '0') * factor; + while(is_octal_digit(c)) { + obstack_1grow(&symbol_obstack, (char) c); next_char(); } + obstack_1grow(&symbol_obstack, '\0'); + char *string = obstack_finish(&symbol_obstack); - if(c == 'e' || c == 'E') { - next_char(); - parse_floatingpoint_exponent(value); - return; + const char *endptr; + lexer_token.type = T_INTEGER; + lexer_token.v.intvalue = parse_int_string(string, &endptr, 8); + if(*endptr != '\0') { + parse_error("octal number literal too long"); } - lexer_token.type = T_FLOATINGPOINT; - lexer_token.v.floatvalue = value; - - parse_floating_suffix(); + obstack_free(&symbol_obstack, string); + parse_integer_suffix(); } static void parse_number_dec(void) { - int value = 0; - + bool is_float = false; while(isdigit(c)) { - value = 10 * value + c - '0'; + obstack_1grow(&symbol_obstack, (char) c); next_char(); } if(c == '.') { + obstack_1grow(&symbol_obstack, '.'); next_char(); - parse_floatingpoint_fract(value); - return; + + while(isdigit(c)) { + obstack_1grow(&symbol_obstack, (char) c); + next_char(); + } + is_float = true; } if(c == 'e' || c == 'E') { + obstack_1grow(&symbol_obstack, 'e'); next_char(); - parse_floatingpoint_exponent(value); - return; + + if(c == '-' || c == '+') { + obstack_1grow(&symbol_obstack, (char) c); + next_char(); + } + + while(isdigit(c)) { + obstack_1grow(&symbol_obstack, (char) c); + next_char(); + } + is_float = true; } - parse_integer_suffix(); - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = value; + obstack_1grow(&symbol_obstack, '\0'); + char *string = obstack_finish(&symbol_obstack); + + if(is_float) { + char *endptr; + lexer_token.type = T_FLOATINGPOINT; + lexer_token.v.floatvalue = strtold(string, &endptr); + + if(*endptr != '\0') { + parse_error("invalid number literal"); + } + + parse_floating_suffix(); + } else { + const char *endptr; + lexer_token.type = T_INTEGER; + lexer_token.v.intvalue = parse_int_string(string, &endptr, 10); + + if(*endptr != '\0') { + parse_error("invalid number literal"); + } + + parse_integer_suffix(); + } + obstack_free(&symbol_obstack, string); } static void parse_number(void) @@ -451,23 +556,17 @@ static void parse_number(void) case '7': parse_number_oct(); break; - case '.': - next_char(); - parse_floatingpoint_fract(0); - break; - case 'e': - case 'E': - parse_floatingpoint_exponent(0); - break; case '8': case '9': next_char(); parse_error("invalid octal number"); lexer_token.type = T_ERROR; return; + case '.': + case 'e': + case 'E': default: - put_back(c); - c = '0'; + obstack_1grow(&symbol_obstack, '0'); parse_number_dec(); return; } @@ -476,11 +575,6 @@ static void parse_number(void) } } -static inline int is_octal_digit(int chr) -{ - return '0' <= chr && chr <= '7'; -} - static int parse_octal_sequence(const int first_digit) { assert(is_octal_digit(first_digit)); @@ -491,7 +585,7 @@ static int parse_octal_sequence(const int first_digit) if (!is_octal_digit(c)) return value; value = 8 * value + c - '0'; next_char(); - return value; + return (char_type)value; } static int parse_hex_sequence(void) @@ -510,7 +604,7 @@ static int parse_hex_sequence(void) next_char(); } - return value; + return (char_type)value; } static int parse_escape_sequence(void) @@ -583,7 +677,7 @@ static void parse_string_literal(void) switch(c) { case '\\': tc = parse_escape_sequence(); - obstack_1grow(&symbol_obstack, tc); + obstack_1grow(&symbol_obstack, (char) tc); break; case EOF: @@ -598,7 +692,7 @@ static void parse_string_literal(void) goto end_of_string; default: - obstack_1grow(&symbol_obstack, c); + obstack_1grow(&symbol_obstack, (char) c); next_char(); break; } @@ -762,7 +856,7 @@ static void parse_line_directive(void) if(pp_token.type != T_INTEGER) { parse_error("expected integer"); } else { - lexer_token.source_position.linenr = pp_token.v.intvalue - 1; + lexer_token.source_position.linenr = (unsigned int)(pp_token.v.intvalue - 1); next_pp_token(); } if(pp_token.type == T_STRING_LITERAL) { @@ -1044,6 +1138,20 @@ newline_found: void init_lexer(void) { strset_init(&stringset); + + type_int = make_atomic_type(ATOMIC_TYPE_INT, TYPE_QUALIFIER_CONST); + type_uint = make_atomic_type(ATOMIC_TYPE_UINT, TYPE_QUALIFIER_CONST); + type_ulong = make_atomic_type(ATOMIC_TYPE_ULONG, TYPE_QUALIFIER_CONST); + type_longlong = make_atomic_type(ATOMIC_TYPE_LONGLONG, + TYPE_QUALIFIER_CONST); + type_ulonglong = make_atomic_type(ATOMIC_TYPE_ULONGLONG, + TYPE_QUALIFIER_CONST); + + type_float = make_atomic_type(ATOMIC_TYPE_FLOAT, TYPE_QUALIFIER_CONST); + type_double = make_atomic_type(ATOMIC_TYPE_DOUBLE, + TYPE_QUALIFIER_CONST); + type_longdouble = make_atomic_type(ATOMIC_TYPE_LONG_DOUBLE, + TYPE_QUALIFIER_CONST); } void lexer_open_stream(FILE *stream, const char *input_name) @@ -1067,7 +1175,7 @@ void exit_lexer(void) static __attribute__((unused)) void dbg_pos(const source_position_t source_position) { - fprintf(stdout, "%s:%d\n", source_position.input_name, + fprintf(stdout, "%s:%u\n", source_position.input_name, source_position.linenr); fflush(stdout); }